After some "more" fiddling with IM 6.6.9-4 Q64 + HDRI
I have stumbled a strange JPG bug ...
Whenever the image is Identified as "8-bit PseudoClass 256c" both "convert" and IMDisplay simply show a "black" image ...
identify dsc1860b.jpg
dsc1860b.jpg JPEG 1288x989 1288x989+0+0 8-bit PseudoClass 256c 176KB 0.047u 0:00.092
In IMDisplay it is completly black ...
Nonetheless if I call "convert -monitor dsc1860b.jpg -distort resize 20% -auto-level tt1.jpg"
Load image[dsc1860b.jpg]: 988 of 989, 100% complete
Level image[dsc1860b.jpg]: 197 of 198, 100% completee
Save image[tt1.jpg]: 197 of 198, 100% complete
I get a "damaged" image like this ...
Using manually compiled Windows version with HDRI quantum 64 - 6.6.9-4
Any ideas ?
JPG Problem + Quantum Range 64
-
- Posts: 88
- Joined: 2010-06-29T14:36:09-07:00
- Authentication code: 8675308
-
- Posts: 88
- Joined: 2010-06-29T14:36:09-07:00
- Authentication code: 8675308
Re: JPG Problem + Quantum Range 64
After some tests I noticed that the previous problem is not appearing if I compile with Q16 or Q32 ...
Still keeping the HDRI flag ON ... ( or at least is what the "convert -version " reports ...
So now the this kind of JPG image ( tested with IMDisplay) is now correct ....
Even calling the "-auto-level" option works correctly and does not give the previous problem ...
Somewhere in the JPG library there is a "problem" with Quantum 64 ...
Strange why are affected the "Pseudo Class 8 bit 256c" JPGs ?
Comments ?
Still keeping the HDRI flag ON ... ( or at least is what the "convert -version " reports ...
So now the this kind of JPG image ( tested with IMDisplay) is now correct ....
Even calling the "-auto-level" option works correctly and does not give the previous problem ...
Somewhere in the JPG library there is a "problem" with Quantum 64 ...
Strange why are affected the "Pseudo Class 8 bit 256c" JPGs ?
Comments ?
Re: JPG Problem + Quantum Range 64
We don't advertise Q64 because it has not yet been validated and will produce incorrect results in some cases. Whenever ImageMagick is compiled with HDRI support, use Q16 as the quantum depth.